Structures Report - Niall Sharples


The site can be divided into three significant and separate areas:

Area 1/5 This was the principal area of excavation on the summit of the hill in both the first (1) and second (5) season. An area was opened up which covered the broch and the external structures surrounding the broch on the north, west and south sides. Excavation immediately to the east and north east side of the broch was restricted by the developers activity.

Area 2 This was an L-shaped area to the north east lying between house plots 1 and 2 and the access road. It was partially explored in the first (contexts 100-126) and second season (contexts 300-320).

Area 3/4/6 This area consists of two test pits excavated in the first season (areas 3 and 4, contexts 127-132), the stratigraphy recorded in the contractors sewage trench (area 6, contexts 144-147, 1076) and a machine cut and two small trenches excavated in the second season (area 3, contexts 400-435). All of these trenches lie on the southern slopes of the hill below the broch.

Within each area the contexts have been divided into blocks and sub blocks defined by there spatial relationships. The blocks do not relate to the phasing of the site and most blocks include features belonging to more than one phase. The subdivisions, however, are exclusively within a single phase. See publication figure 15 for a schematic diagram showing the relationship of all the blocks, subdivisions and phases.

Anything with a zero subdivision is an unstratified part of that block.

Area 1/5

This is divided into ten blocks. These blocks are mostly distinct areas and it is normally not possible to link the stratigraphy between the different areas. The exceptions are blocks 1 and 8. Block 1 represents the recent topsoil and modern disturbance which overlies the whole of the area and block 8 represents the features, old ground surface and natural deposits which underlie the phase 2 and 3 structures.

The principal structure which defines the arrangement of blocks in this area is the broch which lies at its centre. The stratigraphy associated with the construction, use and abandonment of this structure is numbered block 7. Immediately to the north of the broch was an area of deep stratigraphy and important structural remains. The layers, structures and features in this area were labelled block 6. To the north of this block was the Middle Iron Age ditch. The contexts infilling this ditch were separated from the block 6 features by a soil dump which may be the remains of the adjacent rampart. The rampart, ditch and associated contexts were labelled block 5. A large unexcavated baulk separates block 6 from the adjacent contexts to the west. These were labelled block 4. To the south west an area of bedrock separated block 4 from block 3. Similarly block 3 was separated from block 2, to the south, by an area of bedrock. Block 11 and 12 lay in areas excavated in the first season. Block 11 lay in the south quadrant of this area, directly to the south of the broch. Block 12 lay to the south east of the broch separated from block 11 by a baulk.

Block 1

This consists of twenty two contexts:-

1, 2, 3, 4, 5, 7, 20, 21, 91, 94, 95, 98, 200, 201, 202, 203, 204, 236, 244, 245, 246, 1013

This block isolates the topsoil and modern dumps that cover the archaeological stratigraphy.

It is not subdivided.

Block 2

This block isolates a group of contexts in the south west corner of the area excavated in the second season. It is delimited on the east by the front face of the broch wall, on the west by the quarry, to the south by recent disturbance and to the north by an area devoid of archaeological stratigraphy.

2/1 This consists of ten contexts:-

228, 235, 237, 265, 266, 570, 571, 594, 615, 618

These make up a badly built and badly preserved structure probably of Late Iron Age date.

The structure was defined by walls to the east, north and south, the western wall had been removed by the quarry. The eastern wall, 228, survived as about 6 small slabs laid horizontally in an arc. These stones were quite clearly deliberately laid (though the plans are not very convincing), and form the principal reason for the identification of this structure. The south wall, 570, was in contrast defined by a series of edge set slabs which were up to 0.7 m long. These were supported by a rubble bank. The northern wall, 571, was the most difficult to identify and there is no certainty that the rubble planned as a wall is actually in situ wall. Though at least two large horizontal slabs in the south east corner were convincing. At one point in this wall an arrangement of slabs suggested a small cist but excavation of the soil fill, 594, revealed nothing.

The surface on which this structure was built was very uneven. The bedrock was dipping steeply from north east to south west and it is possible that some quarrying of the bedrock occurred just inside the north wall. Consequently the floor of the structure was partially bedrock and partially deliberate make up. A thin layer of charcoal flecked soil, 266, was identified as a floor level and on it were patches of red ash, 615. At the centre of the structure was an arrangement of large stones which gave the impression of being two sides of a stone hearth but there was no extra concentration of charcoal or any other sign of burning at this point. These slabs coincided with a shallow feature filled with a coarse gritty silt, 618, which appeared to be redeposited natural.

Covering the floor and much of the walls was a considerable amount of loose rubble. This varied slightly depending on the size of the stones, the proportion of stone to soil and the stability. Three different contexts were identified 265, 235 and 237. 265 was below 235/237 and was a more compact layer of smaller (but medium sized stones) on the west side of the structure. 235 and 237 were both very loose voided rubble layers with evidence for considerable recent disturbance by burrowing animals. 237 consisted of large sub angular blocks whereas 235 was more slab like medium sized stones. Within this rubble two stones stood out as being of greater significance than most. These were thin slabs over 1 m long and only 0.20 - 0.30 m wide both appear to have originally been vertical uprights projecting from the floor of the house. Other than these much of the stone in these rubble layers appears to have derived from the collapse of the northern and southern walls.

2/2 This consists of one context:-


This is an isolated patch of tipped flat slabs lying directly in front of a preserved section of the broch wall. They seem to represent collapse from the broch wall.

2/3 This consists of two contexts:-

531, 611

These were soil layers, with only a few isolated stones, which were used to level up the ground on the south side of the structure (2/1). The bedrock was dipping steeply to the south and up to 0.4 m of soil, 611, was required to create a level foundation for the south wall. This deposit started about half way across the floor. 531 was deposited in a series of bedrock crevices in the slightly raised east end of the structure. Within this layer was buried a complete vessel (1282). Neither of these layers was fully excavated.

Block 3

This block isolates a group of contexts on the west side of the trench. It is defined to the east, north and west by the edge of the trench and to the south by an area without archaeological stratigraphy.

There are two subdivisions.

3/1 This consists of four contexts:-

205, 206, 209, 210

These are dumps of soil and rubble deposited to infill the area above structure 3/2.

205 and 206 are discrete dumps of rubble in a dark brown silty clay matrix. 206 directly overlay a dark grey silt, 210, which lay on the edge of the hollow of structure 3/2. It in turn overlay a dark brown clay silt which may be redeposited glacial clay, 209.

3/2 This consists of five contexts:-

214, 215, 216, 217, 227

These are the remains of a structure only partially exposed by the recent excavations.

This structure was placed in a hollow quarried into weakened bedrock. The features were exposed planned and recorded but not further excavated. The principal defining feature was a wall, 214, backed onto the bedrock and defining the eastern edge of the structure. A clay layer, 215, lay between 214 and the bedrock and appeared to have acted as a bonding agent to hold the wall in place. The wall was 0.8 m long and stood 0.30 m, 5 courses high.

In front off and lower than this was wall, 216. This again faced north west and was revetted into a clay silt, 217. The wall ran for 1.30 m but was only a single course, 0.13 m, high. In front of this wall was a layer of dark grey clayey silt which may be redeposited natural. Both 217 and 227 may be floors with wall 216 a change in the level of the floor.

In all of these contexts and those of 3/1 finds were rare, fragmented and insignificant.

Block 4

This block isolates a group of contexts in the north west corner of the excavated area downslope from the broch. It is defined to the west and north by the edge of the trench, to the south by the broch wall and to the east by a wide unexcavated baulk.

There are four subdivisions.

4/1 This consists of four contexts:-

218, 288, 508, 515

These are dumps of soil and rubble which overlie the structural features discussed below.

The uppermost layer, 218, was a discrete layer of small to medium sized angular slabs clearly dumped along the southern edge of the area. This overlay a much more widespread layer of clay silt, 288, with a smaller proportion of large angular blocks. Below this the stratigraphy split in two. 508 was a yellow brown mottled clay with charcoal flecks. This overlay the infilled house on the north west corner of the area and is a dump of ash. 515 was another rubble layer concentrated along the southern edge of the area and against the eastern section. A radiocarbon date (GU-2925) of AD80+/-90 was obtained from a cattle leg in this layer. After calibration the date suggests this backfilling occurred by AD390-595 (69% confidence). Amongst the stones of this rubble layer was an apparently deliberate arrangement of very large boulders. But no specific deposits could be associated with this arrangement and its status is unknown.

In general these deposits appear to be a deliberate attempt to infill an inconvenient hollow and to cover an abandoned structure.

4/2 This consist of eight contexts:-

514, 516, 562, 573, 574, 597, 598, 606

These are the remains of a Late Iron Age structure and its associated occupation and destruction deposits.

The structure was found in the north west corner of the trench and most of it lies outside the area excavated. As exposed, it was a 2 m arc of walling, 597, 0.7 m, 3-4 courses, high. Abutting the wall was a carefully laid floor of flat slabs, 606. The wall had been sunk into the underlying natural till, 572, and this partially explains the quality of preservation. No specific occupation levels were identified and the whole structure was filled with a layer of silty clay, 598. This layer rose to be level with the top of the house wall and within it there was no sign of any wall collapse. It would appear therefore that the house had been kept clean until it was abandoned when it was carefully demolished and deliberately infilled. It must be remembered, however, that most of the structure was not excavated.

Around the house and covering the area between the house and the rising slope of bedrock were three thin soil layers 562, 516, 514. 562 gave the appearance of a preserved old ground surface in the west section as it was a thin brown clay loam lying on natural till immediately behind wall 597. The two later layers could also be reasonably interpreted as dumps infilling the earlier ditch as part of the construction of the house. 514 was a distinctive pale olive colour which suggests it was redeposited natural. Unfortunately, however, this relatively simplistic interpretation is spoilt by the fact that 562 overlay wall 597 and the infill 598 at the northern section, thus apparently making all these layers later than the destruction of the house and related to the infilling of the area (see 2/1). Between layer 562 and 516 was an ephemeral scoop or hollow, 573, filled with a gritty loam, 574. This feature is of unknown significance.

4/3 This consists of one context:-


This was a layer of rubble lying in front of the broch wall which is presumed to result from the collapse and destruction of the broch in the Late Iron Age. It is stratigraphically unconnected with the other layers in this block.

4/4 This consists of three contexts:-

490, 551, 561

This was the Middle Iron Age ditch and the soil fill of this ditch.

Fill layer 551 is distinguished from 561 by the presence of large angular boulders and these overlay 561. 551 may be deliberate infilling over a natural soil fill 561 but this is not as clearly deliberate as the other later layers.

The ditch, 490, was shallow and ephemeral at this point. Much of the effect is created by the slope of the bedrock down from the summit of the hill and there is no evidence that this was emphasised by quarrying. The actual ditch is only a 0.5m deep gully at the point where the bedrock slopes under the natural till, 572.

Block 5

This block isolates the contexts in the northern extension of the excavated area. The area is defined to the west and north by the edge of the trench, to the east by the access road and to the south by the bank of the Iron Age ditch. This was an area of deep stratigraphy as most of the contexts lie within the infilling of the Middle Iron Age ditch. Consequently with the exception of a block of contexts lying to the north of the ditch the subdivisions are arranged in stratigraphic order.

There are five subdivisions.

5/1 This consists of 14 contexts:-

220, 221, 232, 243, 264, 518, 527, 529, 536, 537, 541, 542, 544, 546.

These contexts represent activity associated with the historically documented crofting activity of phase 5.

When first cleaned the area gave the appearance of being a very heavily disturbed area of rubble. Two short stretches of wall were clearly visible. 220 lay on the south side of the area and was about 0.6 m wide and 1.4 m long but stood only one to two courses, 0.12 m, high. It could not be associated with any other feature and lay over the fill of a large pit (see below). The subsidence of the fill of this pit may have caused its preservation as the remaining structure would have stood on the ground surface to be disturbed. The other wall, 264, was even more nebulous. It lay on the eastern edge of the area and was 0.66 m wide and 1.75 m long and only 0.10 m high. It was thought to define a rectangular area extending outwith the area to the east but careful excavation could not firmly establish this.

Having expended some effort in cleaning, planning and examining these walls it was decided that this rubble spread was too disturbed to contribute much information. The remaining rubble was removed quickly as layers 243 and 232.

Once the rubble had been removed it became clear that a large figure of eight shaped pit existed in the middle of this area. Excavation revealed that this was two separate pits 546 and 529. 546 was earlier and was roughly 0.75 m deep and approximately 1.7 m in diameter. Its fill was removed as one but three layers were differentiated in section, 542, 544, 518. The later pit, 529, was separated from 546 by a carefully built stone wall, 541, and was approximately 0.9 m deep and 1.9 m in diameter. It was filled with four layers, 541, 537, 536, 527, containing variable amounts of rubble and modern artefacts. The final layer, 221, was a very distinctive sub rectangular patch of rubble which was visible after the initial cleaning of the area and which may have been a hard standing for later activity.

5/2 This consists of 24 contexts:-

224, 249, 250, 252, 502, 505, 506, 520, 521, 522, 523, 524, 525, 526, 530, 538, 545, 548, 550, 564, 565, 568, 585, 602

These represent the final fills of the middle Iron Age ditch and consist of fine silt layers, layers of paving and other occasional features.

The stratigraphy in this subdivision can be divided by the three important silt layers 252, 564, and 249. These layers were all very similar red brown fine clay silts with only a scatter of medium to small stones. 252 was the lowest layer which defines the bottom of the subdivision, 564 separates the lower path from the upper path and 249 marks the final layer deposited in this subdivision. The latter layer was slightly different as it had a higher stone content and patches of red clay, 250, which represent ash dumps. It was also disturbed along the east edge by later animal burrowing, 522, 526. These silt layers would appear to have formed by gradual soil creep into the declivity at the top of ditch from the adjacent slightly higher ground to the south and east. It is just possible, however, that these were deliberately deposited layers placed to infill the hollow of the ditch.

Two major strata of activity are represented within this period of silting. The uppermost contained by far the greater number of features. The principal feature was a bounded layer of large stone slabs, 506, which extended from the southern part of the western section 5.5 m to stop just before the eastern section. This layer was one stone thick and 1.2 m wide and was defined on the west by a line of small upright stones, 505, which projected 0.30 m above the layer of horizontal stones. The northern end appeared to be a deliberate termination as a few smaller stores were used to round of the edge. At the southern end, and unfortunately partially obscured by extension into the unexcavated area, the feature was changing direction or coming to an end.

Apparently contemporary with the paving or slightly later were three post holes 548 (545), 550 (525), 538 (520). 548 actually lay within the eastern edge of the paving, 550 lay beside this just outside the paving whereas 538 was about 1m further to the south. No obvious interpretation can be made of these but 548 and 550 are connected not just by there proximity but by there rectangular shape and proportions; 0.45 by 0.35 m and 0.35 m deep and 0.40 by 0.30 m and 0.15 m deep. Whereas 538 was circular, 0.25 m in diameter, and 0.25 deep. Lying on top of the paving was a small patch of fine silt, 524, which was probably ash from a fire. There was no burning visible on the adjacent stones to indicate this was in situ but this is not conclusive evidence. At the south end the paving was overlain by a pile of large rounded and angular boulders, 523, roughly 2 m by 1.5 m in extent.

Some features on the west side of this paving appear to have been truncated by its creation. These were two arcs of stone slabs, 530 and 521, which might indicate a building. These were only a single stone thick and given the limited extent present they can not be confidently interpreted. Within the inner arc was a soil layer, 502, which was only slightly different from the underlying silt and did not appear to be a floor. Also possibly slightly earlier than this paving was a patch of rubble, 565, adjacent to the east section. This covered a layer of clay with charcoal, 568, which appeared to be a dump of ash. Neither of these features is interpretable above the level of activity roughly contemporary with the later paving.

The lower paving, 585, is, like the upper a band running roughly north to south, approximately 1.3 m wide, with a line of small orthostats defining the east and west edges. Unfortunately it extended into the unexcavated area to the north and east and appears to be terminated by disturbance to the south. So its full extent is unknown. Unlike the upper paving there were no other features associated with this layer. Two soil layers were, however, isolated on either side of the paving which appear to be contemporary with it. On the east side was 224 which is a brown silt similar to the over and under lying silts. It did, however, contain larger quantities of artefacts including large fresh bones. A radiocarbon date, (ut-1654) of ad560+/-30 was obtained from a cattle leg in this layer. This would indicate deposition of this material between AD625 and AD665. On the west side was 602 which was a darker silt with large quantities of small pebbles. This might have been a deliberate deposit laid to create a hard surface at the same level as the paving.

This subdivision can be summarised as a period of time when the depression created by the underlying ditch (5/4) was gradually infilling by natural processes of soil creep. During this period access across this silt filled hollow was required and stone paths were created. Associated with the second of these paths was a variety of other features which slightly precede and succeed it. The path was subsequently covered by a silt layer which covered the area prior to the crofting activity of phase 5.

5/3 This consists of 10 contexts:-

225, 251, 519, 552, 555, 637, 643, 646, 650, 651

These represent the deliberate infilling of the middle Iron Age ditch and the construction of several structures in the Late Iron Age.

This phase of activity begins with the deposition of a clay layer, 651/251, about 0.3 m thick. This acted as a foundation sealing the earlier ditch fills before the construction of a building 643/225 on the east side of the area excavated. Contemporary with this building on the west side of the ditch was a dump of rubble, 650, which was overlain by a silt accumulation, 646, and then a second silt layer, 637, which appears from its colour, form and consistency to be a deliberate dump. Contemporary with this activity was the creation of another structure, 519, which cut into the bank on the south side of the ditch.

The two buildings in this phase were only partially explored. 643/225 largely lies to the east of the area excavated, under the access road into the housing development. Little more than a 1.2 m stretch of the edge of a much more extensive structure was excavated. This consisted of a line of small slabs, 225, which kinked to create a recess built from larger slabs, 643. The recess may represent an alcove built into the wall of the building or possible a point of access, step, into the passage of a building. There was no clearly distinguishable floor level. The later red brown silt simply came down onto the underlying redeposited clay.

The second structure which was better preserved in some respects is just as difficult to interpret. Cut into existing deposits, 787, was a 1.5 m arc of walling, 519. This was well built surviving up to five courses, 0.5 m, high. Unfortunately the structure was completely truncated at both ends by the later, phase 5, pit. No evidence for the structure exists on the west side of this pit. It would therefore have to be a very small structure. Again there was no distinct floor level. The infilling, which consisted of two largely indistinguishable layers of reddish brown clayey silt, 552 and 555, simply rested on the preceding deposits.

5/4 This consists of 14 contexts:-

212, 238, 621, 638, 663, 665, 666, 667, 671, 674, 675, 676, 677, 787

This block describes a large ditch and bank of Middle Iron Age date and the earlier fills of this ditch

Ditch 677 was explored by a slot, 3 m wide at its mid point, cut from the top of the natural till, 671. On the west side of this trench the ditch was 0.95 m deep but on the east side it was only 0.35 m. So longitudinally the ditch drops 0.40 m. This might indicate that there was an entrance somewhere to the east under the road but it is also possible that the depth of the ditch reflects the underlying natural. The deepest area coincided with a area of limestone bedrock that was badly fissured and pocked with solution holes.

The nature of the underlying natural may also have had an effect on the position of the ditch. The bedrock begins to dip noticeably down to the north just before the ditch began and the sides of the ditch are largely cut through natural till. This phenomenon has indeed caused some confusion over the precise edge of the ditch. Two layers 671 and 238/638 had been interpreted as redeposited natural (similar to 651/251 in block 5/3) if this was correct it would imply a wide irregular ditch and/or an earlier ditch which has subsequently been recut. However, the west section would suggest 671 is a natural till cut into by the ditch. 238/638 in the east section is more problematic it would appear to indicate a recut but I believe it might also be a natural layer cut by the ditch.

The fill of the ditch therefore begins with 665 which is soil infilling solution holes in the limestone on the west side of the ditch. Above this was 666 a sandy clay layer concentrated on the north side of the ditch which resulted from primary and probably fast erosion of the till sides of the ditch. This underlies and then laps up against a rubble layer 676 consisting of large rectangular blocks which appear to have been dumped in the ditch. This is followed by another rubble dump, 667, of smaller blocks mixed with a sandy clay apparently similar to natural. A cattle humerus from this layer has provided a radiocarbon date (ut-1655) of 80+/-40bc. This suggests the ditch was created before the period 85BC-AD15. On top of this was another dump of rubble, 663, which was differentiated by the presence of smaller angular slabs. These three dumps create a noticeable mound in the centre of the ditch at the west section and against them sandy silts from erosion of the ditch sides gradually accumulate. On the south side this is 675, on the north side 674. The final layer over these deposits is 621/212 a reddish brown sandy clay which presumably largely derives from natural silting but includes large quantities of animal bones and artefacts.

Immediately to the north of the ditch is deposit brown stony soil, 787, which was cut by a block 5/3 structure, 519. This material was not excavated but it may represent the remnants of a bank adjacent to the ditch.

5/5 This consists of six contexts:-

503, 504, 581, 586, 601, 603

These are contexts identified to the north of the ditch which are largely stratigraphically unconnected with the contexts in the ditch.

The earliest context was an arc of stones, 503, which suggested a possible structure was present. There was, however, little coherence to the stone line and though in some places it appeared to be coursed in plan the feature is not convincing. To the south a soil layer, 603, overlay the natural and this appears to be quite a late accumulation possibly equivalent to the layers accumulating around the upper path of 5/2. This was in turn overlain by 601. A layer thickest to the north east and differentiated from 603 by containing large quantities of rounded pebbles. This was cut by a shallow circular hollow, 586, 1.2 m in diameter and 0.15 m deep filled with medium to small rubble, 581. The final layer was 504, a stony clay silt with modern material presumably contemporary with 5/1.

Block 6

This block isolates a group of contexts between the broch and the external bank and ditch of block 5. It is defined to the north by the Iron Age bank to the east by the access road, to the south by the external face of the broch wall and to the west by a large unexcavated baulk.

There are seven subdivisions.

6/1 This consists of 13 contexts:-

92, 93, 96, 231, 239, 247, 260, 532, 533, 534, 558, 560, 640

These are ephemeral layers and features which may represent a brief final occupation of the site before it was abandoned at the end of phase 3.

The defining features for most of the contexts in this area were two stone lines, 532 and 231. These define a hollow infilled with rubble layer 96, 239, 260. These sealed a patch of orange clay 640/534, which appears to be a hearth and a soil layer, 533, with charcoal and patches of orange clay derived from the hearth. The wall on the south side of the area, 532, was a line of 4- 5 large stones set on edge extending for 2.2 m. The wall on the north side was in contrast represented by a line of three horizontally laid blocks. Vertical stones along the northern edge of the area appear to have been the collapse from this wall rather than a deliberately arranged feature. Cutting diagonally across the area between these two defining walls was a another stone line, 247, five stones set on edge extending for 1.4 m and standing 0.3 m high.

Behind wall 532 and presumably acting as some sort of structural infilling were two rubble layers 558 and 560. 558 was made up of small to medium sized stones many of which were clearly water rolled. This overlay another rubble layer 560 which had a much more significant soil (silty clay) component. This layer and wall 532 overlay 595 which was probably associated with the decay of the block 6/4 structures and the construction of the block 6/2 building. Two soil and rubble layers 92 and 93 were identified in this position when the eastern section was cleaned in the first season but it is unclear exactly what layers these are the same as in the second season.

In general these features could be interpreted as the creation of a defined space (probably not a building) by the construction of the walls 231 and 532 and the associated infilling behind 532. Occupation inside the structure was represented by layers 534 and 533 and the stone line, 247, may have formed part of this occupation. The structure was finally infilled by a layer of rubble 96/239/260 which was deliberately deposited.

6/2 This consists of 28 contexts:-

222, 240, 248, 556, 587, 588, 593, 595, 596, 607, 610, 612, 614, 620, 626, 627, 628, 631, 633, 639, 642, 647, 648, 649, 652, 653, 664, 712

These represent the construction, use and abandonment of a Late Iron Age building.

The structure of the building was defined by a series of walls seldom greater than two courses high and often appearing only as discontinuous lines of individual stones. A lobate structure was defined by three cells extending from a central rectangle. The western edge was particularly badly defined as the stratigraphy was truncated from east to west.

The eastern cell was the best preserved area of the house and consisted of a wall 222 which defined three sides of an area 2 m by 0.9 m. The fourth side, which separated this cell from the central part of the house, was demarcated by a line of five slabs, 588. Some of these were laid flat, and two courses survived, but others were set on edge. Wall 222 was made of medium to large slabs laid horizontally and surviving up to 0.16 m, 3 courses high in places.

The southern cell was defined by walls, 649 and 626. Only the east and part of the south side of this cell survives but the pattern of rubble suggests an area 1.2 m by 0.6 m was defined. The cell was separated from the central part of the house by a line of 4- 5 medium sized blocks, 627. Wall 626 was up to two courses high and made from medium sized flat slabs.

The northern cell was the worst preserved. The most clearly defined stretch of wall was the north east corner where 4 medium sized slabs survived. The edge of the slab layer 600 (see below block 6/5) indicated that an area roughly 1.4 m by 0.7 m was defined. The cell was separated from the interior by a scarp cut into the underlying deposits and an area of horizontal slabs (unnumbered).

At the south west corner of this cell was a rectangular pit, 712, 0.4 by 0.4 in plan and 0.2 m deep. This was lined with flat slabs and infilled with stones and sandy silt, 664. This may well be a tank or container connected with the activity taking place inside the building.

As has already been mentioned the western edge of the structure was more difficult to identify. Wall 629 appears to define a threshold or entrance but it is possible this relates to the structure belonging to block 6/4 as the stratigraphy is very truncated at this point.

The southern edge of this area was defined by wall 628 which continues the line of wall 627. This could be the south wall to an entrance to the structure and it is noticeable that the interior paving, 587, extends to the east as though marking an entrance. Wall 587 comprised 10 stones many of which, contrary to normal practice, had been set with their long axis perpendicular to the line of the wall.

The structure sat on a layer of soil which was subdivided into numerous different layers depending on what part of the structure it underlay. The central area was 612/642, the eastern cell was 639, the southern cell was 614/652, the northern cell was 610, to the east of the eastern cell was 607 and behind the southern wall was 647. This was normally a dark brown fine clay silt with a minimal component of stone (about 25% small and angular). The later two contexts were stonier than average. Charcoal, pottery and bone were abundant but not well preserved in this layer. This layer can be interpreted as a soil deliberately deposited, possibly after earlier structures had been cleared, as a level foundation on which the building could be constructed.

The interior floor of the structure consisted of a layer of large flat slabs, 587, which formed a paving covering an area 2 m by 1 m extending from the centre of the house west towards its edge. This slab floor partially sealed a patch of silty clay, 631, in the centre of the structure with very large quantities of ash and charcoal. This sat inside a sub rectangular arrangement of stones, 633, and appears to be the remains of a primary hearth which was later sealed by the paved floor. A gap in the paving directly above the hearth indicates some form of continuing use of this hearth.

No very clear floor layer could be isolated in this structure the paving and underlying soil layers were sealed by another layer of dark brown silty clay 248/556 which filled the structure and overlay the demolished walls. This soil layer contained some angular slabs and rubble which could have derived from the collapse of the structures but there were remarkably few given how badly preserved the walls were. It seems more likely that the buildings were deliberately dismantled and the area infilled. Sitting amongst the soil infilling the structure was the top of a rotary quern. This had been recorded as coming from 610, which is interpreted as the foundation layer but it is more likely that it came from the infilling or just possibly the use of this house. It demonstrated the difficulty of associating finds with the occupation, the construction and the infilling of the structure when there is no definable floor layer.

To the south of the structure was rubble layer 595 which probably represents the decay of the early structures and perhaps the broch wall. To the east was a jumble of rubble and soil layers 596/593/240 which were difficult to interpret. They were dark brown silty clays with a variable quantity of small to medium sized slabs. These infilled a hollow which may have been dug into the earlier block 6/4 structures to recover good building stone (in particular the ends of 553 which might have been large blocks). They overlay layers 648 and 653 which were equivalent to 620/607 et al and were used to infill respectively the cell defined by wall 630 and the passage between it and wall 624.

6/3 This consists of seven contexts:-

553, 554, 559, 567, 569, 575, 576

This represents the construction and infilling of a small cell to the east of the building in 6/4.

The structure consisted of a relatively well preserved horseshoe shaped wall, 553. This enclosed an area up to 2 m by 1.8 m. The wall stood to a maximum of 6 courses 0.6 m high in the north east corner. It was constructed in a distinctive manner, slabs about 0.3 m by 0.4 m were set on edge as the basal course and above these smaller slabs or blocks were laid horizontally. The ends of the wall were both truncated and it is therefore impossible to definitely relate this structure to the other buildings surrounding the broch. It seems likely, however, that this truncation was due to robbing (discussed in 6/2) and that this building is contemporary with the building in 6/4 (discussed below). It may be the case that very large rectangular blocks similar to these used at the corner of the 624-625 and 630-655 were present and these would have been desirable building stones.

The wall was cut into 641 which is described in block 6/6. The primary fill of the cell was 576 a thin soil layer restricted to the back of the cell, which may have accumulated during the occupation. The finds from this include an elaborately carved and decorated steatite weight. Overlying this and covering most of the southern part of the cell was another silty clay layer 575. This contained some charcoal but finds were in general infrequent and again this probably represents in situ occupation of the structure. Above these were two layers 569 and 567 which have completely infilled the structure and which probably represent the deliberate infilling of the cell. These were dark brown silty clays. 569 lay at the back of the cell and was distinguished from 567 by having much less stone, a greater quantity of charcoal and large well preserved animal bones. The latter included two complete scapula laid side by side. Overlying this infilling was a thin clay silt layer, 559, with a small proportion of small rounded stones which was in turn sealed by a thin soil layer 554. The latter contained a number of large slabs which pitched into the structure and which appear to be the collapse of the surviving walling of the cell. Both these later layers could be contemporary with the 6/2 activity.

6/4 This consists of 20 contexts:-

617, 625, 629, 655, 668, 669, 693, 694, 698, 699, 700, 701, 702, 703, 706, 707, 710, 711, 739, 789

These represent the construction, use and abandonment of a Late Iron Age building.

The building was defined on its east side by wall 625 which was a horseshoe shaped revetment wall defining a cell 1.2 m by 0.8 m. This wall stood a maximum five courses, 0.4 m, high and was built from medium to large sized blocks and slabs. The wall in the north west corner of this cell had been destroyed by the construction of the later, 6/2, house and this had also removed all of the northern wall of this structure. The south west corner of the cell was marked by a very large orthostatic block. This complemented the large blocks of stone at the ends of wall 630 on the other side of the passage at this point.

The south side of the building was defined by a revetment wall 655. This was part of a continuous wall which appears to follow the edge of the broch in a series of arcs which at their closest point were 0.75 m from the basal course of the broch revetment. It seems likely that this wall precedes the construction of the house and that it was modified as part of the house construction. A 2 m stretch of wall survived and this was made from medium sized slabs and blocks and survived up to four courses, 0.50 m, high. The east end abutted a large orthostatic block which partners that for wall 625 and cell 630. The west end was truncated by more recent disturbance. However, a feature, 694, 0.35 m in diameter and 0.12 m deep with a stone free silt fill, 693, lay in front of this end of the wall. This is the right size to be the position of another orthostatic block defining a cell about 1.6 m by 0.6 m.

As has already been noted all the other walls of this building have been robbed. A possible contender for the western wall of the structure was located early in the excavations. 629 was a line of six slabs which belong to the later 6/2 structure. Reanalysis of the stratigraphy, however, suggests that this was in fact an earlier feature and it is possible that this was an isolated part of the structure described here reused in the later building. This wall was backed by soil and rubble layers 617 and 669 which may be part of its construction.

Leading from the two large blocks in the south east corner of this building is a passage and cell described as block 6/6.

The interior of this building had, unlike the other structures in this area, a floor level with a central hearth and other features. Like the other structures, however, very little occupation material had survived in situ. The floor was largely the top of sandy silt layer 706 though natural till and bedrock was also used. This layer was only c 0.10 m thick at its maximum depth and in places it underlay the east wall, 625. The upper surface was very compacted and hard when first encountered. To the north the building cut into a more stony sandy soil layer, 710, and the absent north wall appeared to be backed onto another stony soil layer, 711.

In the centre of the house was a hearth. This comprised a shallow scoop cut into 706 filled with a layer of gritty silt with small stones, 739, and a paving consisting of four horizontal slabs, 789. These were covered by a black charcoal rich silt, 707, which spread out from the hollow to cover a rectangular area, 0.7 m by 0.6 m.

Four features were identified cutting into the floor. One 694 has already been discussed above. 699 and 701 were also associated with this south recess and though not running precisely parallel with the south wall they are roughly positioned close to a line drawn along the boundary between the recess and the interior of the house. These features were irregular in shape on average 0.25 m in diameter and up to 0.15 m deep. Both contained pitched slabs in their packing (698, 700) and were almost certainly post as opposed to stone holes. The other feature, 703, was north east of the hearth and lay close to the south east corner of the east recess. It was a larger, about 0.36 m by 0.31 m, feature and could possibly be the position of a corner stone as was 694. There were certainly no clear pitched packing stones in the fill, 702, to suggest a post hole.

Sealing the floor of this structure was a silty clay, 668, which contained a large proportion of flat slabs. The slabs were medium to large in size and were noticeably laid horizontal. At first it was thought that these stones represented a paved floor but this interpretation had to be discarded because this was not a single layer of slabs, several lay on top of each other, there was no attempt to completely cover the interior or part of the interior, and the soil of this layer surrounded the slabs and separated them from the floor. Within this layer and lying just above the floor were several important finds. Including a steatite lamp, 4285, and a steatite bar and disc mould. Both these objects were complete and undamaged. It is arguable that they were in fact unused. All of these factors suggest that this was a layer deposited to deliberately infill this structure.

6/5 This consists of nine contexts:-

599, 600, 604, 605, 678, 733, 734, 735, 736

This is a group of soil and rubble layers in the north west corner of this area which are difficult to link stratigraphically to the structures in the rest of the trench.

The subdivision can be split into three; 604, 599, 605 and 600 were probably associated with the construction and use of the block 6/2 house, whereas 678 was probably contemporary with the construction of the 6/4 house and 734, 735 and 736 infill the ditch.

The area was not fully excavated and the lowest layer 736 could be well above the base of the ditch. This is a fine sandy silt with very little stone. It is sealed by 735 a much more stony layer of sandy silt. 733 and 734 are very similar, the later having small quantities of charcoal and ash. Both have fewer smaller stones than 735. The overlying layer 678 is a brown soil layer which can be equated with the soil layer 211/634/691/641 associated with the construction of the house 6/4 and cell 6/3 and is one of the few stratigraphic links with the main part of this area.

The principal feature of the upper layers was 600. This is a layer of large blocks and slabs which defines the north west corner of structure 6/2. It does not appear to be a built wall but it may be all that is left of the foundations or destruction of that wall. It isolates the soil layers 605, 599 and 604. These vary in soil colour and rubble content but do not seem to represent any significant structural activity. They may be partly natural and partly deliberate levelling or terracing up of the area over the infilled ditch.

6/6 This consists of 16 contexts:-

211, 566, 622, 623, 624, 630, 634, 636, 641, 672, 673, 691, 705, 709, 724, 725

These contexts represent the construction use and abandonment of the passage linking the buildings of 6/4 to cell 6/3.

A 1.5 m section of the northern wall of the passage was preserved. This can be split in two by a vertical orthostat set perpendicular to the line of the wall. To the west wall 624 was made from medium to large stone blocks and stood 3 courses, 0.4 m, high. Wall, 623, to the east, was much less well preserved. Starting about 0.3 m high the wall soon deteriorated to a single course of small slabs before disappearing altogether. The disappearance is probably related to robbing stone in this area (see discussion in 6/2 and 6/3).

These walls (and walls 625 of 6/4 and 553 of 6/3) were backed onto a brown soil layer variously named 211 (in section and earlier road side cut), 691 (between walls 553 and 625), 634 (?? to the north of the house) and 641. This soil also acted as the foundation and therefore the floor of the cell defined by wall 553 thus raising the possibility that this cell was later than the building defined by walls 623, 624 and 625. For this to be the case wall 553 would have to have been dug out of an existing soil dump which was deposited as wall 623/4/5 was being constructed. It seems more likely that the difference simply reflects the presence of a large hollow (discussed as 6/7) under 553 which had to be infilled. This soil layer was a brown silty clay with many small stones and grits but very few medium to large stones. It can in all the areas be interpreted as a dump deliberately deposited as part of the construction of the extra mural buildings. Large quantities of bone were recovered from this extensive layer and it also contains many other significant artefacts.

The south wall of the passage consisted of two shallow curves of walling continuing the line of wall 655. The eastern curve was 566 and this extended from the truncated section by the road for 1.8 m to a small orthostat which was set into the wall perpendicular to the passage. This wall stood to a maximum of 10 courses, 0.8 m, high at the truncated section but rapidly diminished to the west where in places it was only two courses high. The good preservation adjacent to the section may be slightly misleading as the wall was reused in this area as part of the structure associated with block 6/1 and it is possible (if not likely) that it was rebuilt at this point. This wall was continued from the transverse orthostat to wall 655 by another badly preserved section of walling, 636, seldom more than two courses high.

The area between these walls and the external face of the broch, 208, had been infilled by soil and rubble layers 705 and 724. The former was a compact silty clay which extended behind wall 655 and had a large proportion of small angular stones. The later, 724, lay behind wall 566 and contained a higher density of larger stones. It is possible, however, that 724 was deposited as part of the later remodelling.

The south wall of the passage appears to be particularly badly preserved because it was significantly remodelled. A small circular cell was created on the south side of the west end of the passage by the construction of wall 630 over wall 636. This wall began at the large orthostatic block which marks the east end of wall 655 and ends at another large orthostatic block directly opposite the block which marks the south end of wall 625. The former block is almost certainly part of the reconstruction of wall 630 as behind it wall 655 is bonded into wall 636. This creates a cell about 0.9 m by 0.8 m which has an entrance 0.52 m wide into the south east corner of the building of block 6/4.

The presence of three orthostatic pillars clearly indicate that this cell was built as an integral part of the house described in 6/4 particularly as the northern most of the three is part of wall 624/5. This raises certain interpretational problems. To the east of wall 630 there is no revetment wall linking the large block to wall 636. All that was present was a dump of loose rubble and soil, 673, which was the constructional backing to wall 630. This sat on 725/622 a compacted layer of redeposited natural not dissimilar to 706 which appeared to act as the floor to the passage. If wall 630 was built at the same time as 625/4/3 it would indicate that this wall and the structure was later than the south wall.

The later problem is not too serious as it is perfectly plausible to suggest that the southern wall was built first because this acts as a revetment to the decaying (demolished) broch wall and may even be contemporary with its occupation. Only later was the decision taken to create building 6/4 and this involved a remodelling of this earlier wall. The absence of any wall to cover the back of wall 630 is more difficult to explain. It was possible but unlikely that this was removed as part of the robbing already mentioned, and there is a distinct kink to 636 which it could have extended to. Perhaps the wall was unfinished perhaps they didn't care that it looked untidy or perhaps it was dismantled at the end of the occupation.

The only occupation in the passage was 709 a charcoal rich lens of gritty soil which lay just in front of the cell. This was probably helped to form by a depression in the floor created by the subsidence of the fill of an underlying feature. The passage was then infilled by 668 (discussed above 6/4) and clay layer 672.

6/7 This consists of 19 contexts:-

97, 219, 226, 229, 233, 234, 241, 242, 708, 723, 727, 728, 729, 730, 746, 768, 790, 792, 1055

The consists of layers and features earlier than house 6/4.

The main feature of this subdivision is a large hollow, 233, directly in front of the broch wall. This was 5.5 m across, 1 m deep and at least 3.4 m long. Only the south west quadrant was excavated. The north west quadrant was unexcavated due to constraints of time and the east half extended to the east under the access road There is some indication that this feature originally had a revetted wall as four to five facing stones, 790, were found at the south west corner.

The limestone rock at the base of the hollow was pitted with solution holes and two of these had soil fills, 768 and 723, which contained artefactual material. The first layer that completely covered the base of the hollow was rubble layer, 746. Many of the stones of this layer were large blocks and slabs which would have made ideal building stone. Within this rubble layer were a number of very large stone blocks which appeared to be part of a deliberately created setting, 234. Approximately four stones formed a right angle but the two walls could not be continued for any length in the area examined. Associated with this setting of stones were two discrete lenses; one of gritty clay 241/97 which appears to be the result of decaying stone and another of clay with charcoal, 730, which contained a number of important finds including a whalebone sword pommel and a string of bone and stone beads. The primary rubble layer was sealed by rubble layer 708/226/219/229/242/1055 which contained notably fewer and smaller stones in a silty clay matrix. These layers were rich in finds which included several large pieces of worked whalebone. A radiocarbon date (GU-2924) of ad220+/-70 was obtained from an articulated cattle leg in this layer. After calibration this date suggests the finds were deposited between AD220-410. Both this layer and the underlying layer 746 had noticeably fewer stones and were difficult to separate at the western edge of the hollow. Consequently there the soil was removed as context 729.

To the south west of this hollow is another smaller hollow, 792, rectangular 1 m by 1.2 m and 0.3 to 0.4 m deep. It was filled with a soil layer, 728, containing a large number of medium to large flat slabs. There is no indication of its purpose. It would have lain directly underneath the passage of block 6/6.

The only other context in this subdivision was a rubble layer, 727, in a natural crevice directly in front of the broch wall and sealed by the backing layer, 705, to wall 655 (see above 6/6). This contained a high proportion of large angular stones and may be associated with the decay or even construction of the broch wall.

Block 7

This block isolates those contexts associated with the construction, use, reuse and abandonment of the broch. They are defined by the outer face of the broch wall except to the east where the stratigraphy was completely removed by the access road.

Originally the stratigraphy was divided into twelve subdivisions. These were groups of contexts associated by their location and their contextual relationships. These subdivisions did not, however, create a coherent stratigraphic sequence for the construction and occupation of this block. As this is felt to be more useful to specialist analysis of the site the stratigraphy has subsequently been divided into periods of occupation on the basis of a structural analysis of the contexts. These will be discussed below.

The major problem in understanding this structure was created by excavating in two separate seasons. Because all trace of the remains was removed between the two seasons the only relationship between the two is derived from the site drawings. These are also difficult to tie together as the first seasons grid could not be re-established and was in any respect a theoretical construction as all planning was done from arbitrary lines located by EDM.

The problem is manifest in the shape of the broch. The plan of the broch wall in the second season is circular with a deviation of only 0.10 m from an internal diameter of 7.6 m and 0.16 m from an external diameter of 18.8 m. If the first season's excavations are added on as originally grided up then the circumference of the inner revetment deviates by over 1.0 m and the outer revetment by over 1.5 m. It seems unlikely that this deviation is a true reflection of the original shape of the broch wall as the largest single exposure is the area examined in the second season and this is the most circular. The wall in the first season was planned as four completely separate drawings which were then linked. Consequently it will be assumed that the broch is circular with an internal diameter of 7.6 m and an external diameter of 18.8 m and the first seasons drawings will be adjusted to meet these dimensions.

7/1 This consists of 46 contexts:-

6, 10, 11, 12, 35, 80, 81, 87, 90, 207, 253, 255, 256, 263, 268, 269, 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 500, 501, 507, 509, 510, 1008

These are the soil and rubble layers which overlie the abandoned broch. Though 46 contexts are isolated only four principal features are present. The other contexts are duplicate numbers.

This material is largely of phase 3 but inserted into it was a phase 4 inhumation. The inhumation, 253, must be the latest feature and was presumably placed in a grave, 255, which was subsequently infilled, 256. Unfortunately neither the grave nor the fill could be differentiated in the surrounding rubble layer.

The principal feature of this subdivision is a rubble layer 6/35/80/207/268. This layer completely covered the interior of the broch and most of the broch wall. Given the large area covered it was decided in the second season of excavation to subdivide this layer into 2 m square units as this would give some indication of the spatial variation within the layer. (The subdivisions are 274, 275, 276, 277, 278, 279, 280, 281, 282, 283, 284, 285, 286, 287, 290, 291, 292, 293, 294, 295, 296, 297, 298, 299, 500, 501, 509.) The layer varies considerably in thickness but in the general it was thickest over the interior of the broch, 1.0 m, and thinned to the edge, petering out normally just before the outer face of the broch wall. There was a significant visible depression over the interior and to the north west which directly reflects the presence and absence of the underlying structural remains. The layer was remarkably uniform consisting of about 70% small to medium sized angular slabs. Occasionally large and very large slabs (ie 1008) were present and one of these is clearly seen floating in this layer in section (fig 13). What soil there was was very loose and consequently it was very difficult, indeed impossible, to detect disturbance, such as the grave mentioned above, unless an imported fill was present.

The bottom of this layer was differentiated by a more significant and slightly lighter brown soil matrix. This was a gradual transition and it is not thought to have a great deal of significance. In places, however, a separate context number was applied, 10/11/81/87/90/263/269.

Both these layers are interpreted as the result of the destruction of the broch. Large stones have been removed until all that was left was the smaller stones and stones that were decaying. These through subsequent frost action have decayed down to the present homogeneous layer. The finds from this layer were frequent and are almost exclusively Late Iron Age / Pictish. It seems likely, therefore, that the destruction took place during this period. There was no evidence for Early medieval/Norse robbing and by phase 4 the presence of an undisturbed burial suggests that the period of destruction was well over. The material recovered from this layer includes a large assemblage of small mammal, bird and fish bones which seem likely to be the result of animals occupying the decaying and demolished broch.

Underlying this rubble layer was another rubble layer, 12/507/510. This was, however, very different in character and was restricted to the interior of the broch (fig 16.1). It was composed of large angular slabs and blocks with only small quantities of small angular stones. There was also a much more significant soil component of dark yellowish brown clayey silt. This layer was thickest over and in front of the interior walls of the broch and contained substantial quantities of artefacts, ash and charcoal. It is likely to represent the decay and collapse of the structures belonging to the final occupation of the broch.

7/2 This consists of 34 contexts:-

8, 15, 69, 71, 258, 511, 539, 540, 543, 547, 557, 563, 577, 583, 584, 590, 591, 613, 644, 645, 681, 683, 685, 686, 713, 715, 717, 718, 745, 765, 766, 793, 1017, 1047

These comprise the remains of the construction, occupation and abandonment of the final structure inside the broch wall.

The principal feature of this occupation was the construction of a wall defining an area of the interior of the broch (fig 17.1). Three distinct and separate stretches of this wall were identified; 540 on the east side, 511 on the north and north west side and 590 on the south west side. Though certain features belonging to this phase have been tentatively identified in the area defined by these walls much of the occupation appears to be directly associated with structural features of earlier periods of activity.

The wall does not appear to continue into the area excavated in the first season and it appears that the southern part of the broch interior was infilled during this period. This infilling is represented by layers 645 and 644. 645 was a brown silty clay with many small stones, 644 a much more rubbly layer.

Wall 590 began at the southern section of the second season excavations adjacent to the west end of the slab partition, 580 (part of 7/3). At the section the wall was 5 courses, 0.75 m, high but it very quickly deteriorates and could be followed only 0.9 m to the north of the section. This wall was set into a cut which truncates the red ash layer 680 and also the underlying natural boulder clay.

The line between wall 590 and 511 was marked by the continuation of this cut and associated with it were several post holes two of which could be associated with construction of this wall 717 and 718. Both of these post holes had large packing stones in their fill (685 and 686) which were clearly visible from the floor level. 718 was probably a two post hole with the east post replaced by the western part. It was 0.7 m by 0.4 m wide and up to 0.6 m deep. 717 was a similar size 0.6 m deep and 0.36 m in diameter.

Wall 511 was 3 m long and more or less followed the curve of the original broch wall about 1.7 m inside it. This was the best preserved section of walling in this structure. It was, however, still an extremely badly built construction. The best preserved section stood 6-7 courses, 0.5 m, high. It was constructed of medium to large slabs. One large slab was set on end as an orthostat and a pillar of bonded courses projected 0.2 m into the interior of the building.

Immediately adjacent to this pillar and directly in front of the wall was another post hole, 715. The large packing stones in the fill, 683, of this post hole clearly stood up above the floor level and a patch of very loose soil marked the post pipe. It was 0.29 m by 0.49 m but only 0.15 m deep (noticeably shallower than the supposedly similar post holes 717 and 718). Behind the wall was a rubble layer 539 which had clearly been dumped to infill the area between wall 540 and wall 513.

Wall 511 terminated in an area of loose tipped slabs which clearly derive from the collapse of this wall and adjacent wall 540. Wall 540 unlike both 511 was not an arc following the broch wall but a straight section of wall which joined wall 511 with the end of upright slab 580, though it was unfortunately robbed/destroyed before it reached these slabs. The basal course of this was faced by medium sized angular slabs. Behind it, however, there was a secondary wall face of very large slabs. These large slabs then extended over the basal course to form the upper course of the wall. On top of the basal course was a human skull, 5380. This was crushed almost beyond recognition. As this wall extended across the interior of the broch as opposed to following the edge it had to have a more substantial backing layer. This was done by dumping soil excavated from the interior, layers 613 and 258. The former was a charcoal rich silt layer, the latter a red silt with extensive charcoal flecks which is directly comparable to thick primary deposit 680 on the west side of the broch (see 7/5). This layer is thought to be secondary because it overlay a slab associated with the secondary rebuild of wall 513 which in turn overlay another red silt layer directly equivalent to 680. The layer was, however, fairly homogeneous and it is likely that most of the finds derive from the primary occupation of the broch.

Immediately in front of the northern half of this wall, 540, was a gully 713. It was 1 m long, 0.40 m wide and 0.20 m deep with a shallow U or sometimes V shaped section. It was filled with pitched stones, 681, which projected above the floor. This would suggest it is likely to be the foundation for a stone slab (or timber?) screen. It is likely that this gully preceded this period of activity and was a partition in the interior of an earlier structure. However, it dictated the alignment of wall 540 and was probably dismantled during this period of activity.

The interior of this structure involves the reuse of many earlier features. Notably the central hearth and orthostat 580. The final charcoal rich layer in the hearth, 591, was probably deposited during this period and possibly some earlier layers also. The only definite features which can be placed with this period are pits 584 and 793 and post hole 765/766, as well as the post holes mentioned above.

Feature 584 was a well built stone lined pit lying just to the south west of the hearth directly in front of 580. It was 0.60 m by 0.85 m and 0.60 m deep with a paved floor and stone lining, 745. The lining used rounded medium sized blocks on the north side and large angular slabs set on edge on the east side. It was filled with a silty clay layer, 577, with a fair proportion of angular stones laid in the upper levels. Artefacts were relatively common in this fill and included a fine parallelopiped bone dice, 3563.

Feature 793 was a shallow circular pit to the north of the hearth. It was 0.6 m in diameter and 0.2 m deep and was filled with small to medium sized rounded pebbles, 583. Immediately to the north was post hole 765. This was 0.25 m in diameter and up to 0.20 m deep and had a fill, 766, containing some notable angular packing slabs..

These features were sealed by a clay silt layer 8/15/543/563/557 which completely covered the interior of the structure. This layer was mixed with a fair amount of rubble from the collapse of the structures and could not in general be separated from the decay of the buildings even though it must to a certain extent represent occupation in this period. At the base of this layer lying on the paving of the broch floor were a pair of articulated cattle legs, 547. These were the left rear legs of two separate animals. A radiocarbon date (GU-2926) of ad620+/-70 was obtained from these legs indicating abandonment of the broch before AD615-710 (69%).

7/3 This consists of 41 contexts:-

19, 28, 67, 82, 259, 513, 549, 578, 579, 580, 592, 608, 609, 616, 635, 670, 679, 684, 688, 716, 726, 731, 732, 741, 742, 750, 751, 758, 759, 762, 763, 769, 770, 1007, 1010, 1011, 1015, 1016, 1017, 1018, 1052

This group of features provides the main secondary occupation of the broch and include the most visible internal furnishings of the broch interior.

The principal structural feature is revetment wall 513/1052 (fig 17.2). This was a relatively well preserved wall standing up to 6- 7 courses, 0.8 m high. This was constructed with a basal course of very large slabs set on edge longitudinally. Above these the wall consisted of medium to large slabs and blocks sometimes choked with small stones. At the north end the wall is displaced back, 0.4 m, about 1.8 m before the wall was truncated by the access road.

The wall was identified in the northern quadrant of the area excavated in the first season where it was numbered 1010. Incorporated into this wall was a large upright which turned out to be a complete reused trough. In the east quadrant this wall was slightly more difficult to identify. It probably continued beneath the baulk to the end of wall 1015 where it stopped. 1015 was a radial partition projecting from the face of 1007 and with adjacent radial wall 1016 defined a cell. Both of these piers and wall 1007 were probably part of phase 7/4 as they are similar to a pier projecting from wall 273 which was overlain by wall 513. The south east piers appear, however, to have been incorporated into the structure of 7/3 unlike that on the west side.

This wall backed onto layer 28, 635, 1011, 1017, 1018 which was a dump of soil and rubble used to infill the gap between this wall and the original broch wall.

This wall would have defined an area 6.2 m in diameter an enclosed space of 379 sq. m. A reduction of about 190 sq. m on the original floor area of the broch. The principal feature of the interior was a large hearth which sat in the centre of the broch. This hearth is dated to this period because in places the hearth stones 579 overlay red ash layer 747 and pit 767. This would indicate it is not primary and if derived from 680 (though heavily disturbed) it must follow the end of 7/5. The hearth is edged by large slabs which define an area 0.90 m by 1.10 m. These stones are set flat on the underlying soil layer on the north, east and south sides but on the west side two slabs set on edge were used. None of these stones was carefully positioned and several smaller slabs and blocks were used to fill in the gaps and increase the width of the kerb.

The hearth was filled with three layers related to its use. One of these layers has been discussed above as it seems possible that it belongs to the final use of the interior. The lower layer, 679, is by far the most substantial of the hearth fills. It covered the interior and was approximately 0.13 m thick. This was a fine clay silt with a distinctive reddish yellow colour which must indicate it is an ash layer. Charcoal was common and apart from a large slab lying in the centre stones were small rounded and rare. Over this layer was 592 a thin ash lens.

To the south of the hearth was a line of orthostats, 580, which ran from the south east corner of the hearth to the junction between the first and second seasons excavation. There was no sign of a continuation of this line in the first season's excavation or to the east of the hearth. There were two large split slabs, 2.7 m long, standing 0.42 m above the floor level of the broch. These slabs ran across the edge of pit 660 which would have had to be infilled before it could be put into position. Thus indicating that this division was not a primary feature of the broch construction. The final fills of this pit were contexts 67, 549 and 732. Both had large quantities of red ash; 549 lay on the south side of the stone and 732 on the north side. They were closely similar to 747 which underlay the hearth. these layers must have been in place to support the orthostats and 549 in particular contained a number of medium sized stones which would have given it stability. These contexts probably are the result of disturbance to an existing layer which derived from the conflagration which created layer 680 at the end of 7/5.

There are indications that another slab partition extended from the north east corner of the hearth. This only survives as an irregular gully, 713, containing upright packing stones, 681. Any orthostats present were removed by the later occupation (see above 7/2). The gully appeared to extend up to wall 513.

One other upright division can be dated to this period. This was slab 688 which lay about a metre to the north west of the hearth. The slab was 0.40 high and was originally 1.2 m long. It lay in a gully, 751, 1 m long, 0.40 m wide and 0.18 m deep filled with dark brown silty clay, 750. A groove for the stone had actually been cut through a slab in primary pit, 783. When found the central part of the slab had been removed by pit 793 and the upper part of the stone was found broken off and lying flat adjacent to the west side.

Several post holes have been placed in this period. Three of these post holes, 758, 741 and 731 fall in an arc about 0.7 m in front of wall 513. These post holes are linked by the use of medium sized packing slabs set on edge, 759, 742, 726. This packing was not placed to support the post sitting in the hole as it completely fills the hole. It is best explained as a desire to infill an inconvenient hole after deliberate removal of the posts. These holes directly underlay the wall of the 7/2 structure and if the construction of this wall occurred while the structural features of 6/3 survived then it is likely that the structure would have to be dismantled and the holes infilled.

Two other post holes, 762 and 769, are related by their small size (762 was 0.25 m in diameter and 0.15 m deep, 769 was 0.15 diameter and 0.10 m deep) and the presence of a small number of upright packing stones and dark greyish brown fill 763 and 770. They cut natural and were not sealed by wall 590 and therefore can not be dated with any accuracy. They are distinctly skewed in relation to the concentric line of the wall and appear in an intrusive position in relation to post holes 741 and 758. This would appear to make them pre date 7/3 but the absence of a red ash fill suggests they were either finished before the broch was burnt down in 7/5 or occur in transitional period 7/4.

A possibility also exists that post holes 716 and 69 was in use in the period. 69 was filled with 71 and packing stones 1047. 716 was filled with 684.

Unlike period 7/2 and 7/4 this occupation can be associated with an in situ floor level. This comprised a thin discontinuous charcoal rich silty clay which occurred in four distinct areas. 19 in the area excavated in the first season, 616, under wall 511, 578, around pit 584, and 259, to the west of the hearth and around gully 713. Two discrete patches of ash, 608, 609) were found adjacent to the south west corner of the hearth and a brown soil layer, 670, underlay 259. To the east of this gully this charcoal layer was sealed by a number of flat slabs which included one thin slab 0.73 m by 0.58 m placed directly in front of wall 513. This might have been placed in position because of the presence of underlying pit 743. The slabs may have been required on the east side because the underlying surface in this area was bedrock. The glacial till on the west side was absent.

7/4 This consists of thirteen contexts:-

(9, 14,) 29, 273, 654, 692, 695, 696, 697, 1007, 1011, 1015, 1016

These contexts represent the unfinished reconstruction of the broch after it had been burnt down.

The principal structural feature is revetment wall 273/1011/1007. This survived as a single course of stones abutting the primary wall face and overlain by the stones of wall 513/1010. The stones used were large platy slabs and the height of wall was roughly 0.4 m. The wall had three piers projecting into the interior. The remains of the western pier projected below wall 513 and had clearly been dismantled for the construction of this later wall. The two southern piers, 1015 and 1016, however, had been partially retained for use in phase 7/3 to define an alcove. Between this wall and the primary wall was a silty clay layer, 29/654/1018, with many small to medium sized angular stones and possibly rubble layers 9 and 14.

The very bad preservation of this wall suggests that it was either never completed or it had to be completely dismantled before the construction described as phase 7/3. The former hypothesis is the one most favoured as there are few features which can be definitely associated with it. The only definite examples are a pair of postholes 695 and 696, filled with 692 and 697 respectively, which cut layer 680. One of these, however, was overlain by the wall 273 and so cannot be directly contemporary but it is rather part of the immediately post destruction activity. Both were small about 0.15 m in diameter and 0.10 m deep with red brown peat ash fills, 692 and 697.

7/5 This consists of 43 contexts:-

16, 26, 27, 34, 42, 45, 46, 51, 52, 55, 70, 85, 261, 262, 619, 656, 657, 658, 659, 660, 661, 662, 680, 682, 687, 704, 714, 719, 720, 722, 737, 738, 740, 743, 744, 747, 748, 749. 752, 754, 755, 756, 757, 760, 761, 764, 767, 771, 772, 773, 774, 775, 776, 777, 778, 779, 780, 781, 782, 783, 784, 785, 786, 1032

These contexts represent the primary occupation of the broch interior.

The principal defining feature of this period the inner wall face of the broch, is discussed in subdivision 7/8. For the purposes of this discussion the features can be split into those layers and contexts adjacent to the broch wall and hence protected by the construction of the later features and those in the centre of the broch which survived only when they were protected by a later feature or cut into the ground surface (fig 18.2).

The primary features adjacent to the broch wall were a series of shallow bays defined by ridges of natural soil. Three bays 786, 784 and 785 were defined from the southern edge of the second season excavations and a fourth bay is created by pit 743. There is also some indication that similar bays existed in the area of the first season of excavation but these were not so clearly defined. The bays were approximately 2.2 m long and 1.4 m wide and were sunk about 0.1 m below the bottom of the inner broch wall. These features are likely to have been created by the repeated cleaning of cells defined by some form of radial partition which was either removed or destroyed by the fire that destroyed the broch.

At the inner edge of the partition ridges were 3- 5 post holes which held the principal posts of the broch roof.

Post hole 714 lay between pit 743 and bay 785. It was 0.44 m in diameter and 0.15 m deep and had a fill, 682, of red brown silty clay with several large slabs used as packing stones.

Post hole 761 lay between bay 785 and 784. It was 0.4 m in diameter and 0.3 m deep and clearly visible in the fill, 760, was a post pipe 0.2 m in diameter, filled with ash, charcoal and redeposited mineral soil.

Post hole 719 lay between bay 784 and 786. It was 0.3 m in diameter and 0.3 m deep. This had a silty clay soil, 722, surrounding the in situ packing stones. A possible post pipe was defined by charcoal rich silty clay, 687, and this was sealed by a lens of charcoal and red clay, 720.

Post holes were not identified in the first season of excavation possibly due to the restricted space examined.

The bays varied quite considerably in their fill. 786 had no discrete fill. 784 contained a lens of charcoal rich silt, 764, 0.06 m thick, which covered an area 1.3 m by 1.1 m in the centre of the bay. On the northern edge of this bay was a shallow scoop, 771, 0.50 m in diameter and 0.10 m deep. It had a very distinctive green clay lining, 775, and was filled with a brown clayey silt, 772. Bay 785 had a brown soil fill, 778, and in the south east corner a pit 776, 0.35 - 0.40 m in diameter, 0.15 m deep with a green clay fill, 777. Unfortunately only the south half of this bay could be excavated. As has already been mentioned the fourth bay was a large pit, 743, 2.06 m long and 1.43 m wide and 0.45 m deep. The primary fill of this pit was a grey brown clayey silt, 752, though excavated as one layer 0.25 m thick this contained a number of discrete lenses at its base. This was covered by an olive green silty clay 744 which contained a jumble of large and medium sized stones. These two layers more or less completely filled the pit but two thin brown silts 737 and 738 completed the fill. The former higher layer was intermixed with red clay lenses.

In the area of the first seasons excavation a series of thin layers are probably equivalent to these hollow fills. The primary layer was an extensive but thin spread of green clay 84/46/27. This may be a continuation of layer 661 in pit 660 which acted as a floor over the bedrock where the natural till had been removed. Above this was a more restricted rectangular layer of charcoal, 45, which may be directly equivalent to 764.

These bays were sealed by several more extensive layers. The first was 740 which covered 784 and 785 and occurred as layer 42 in the south east quadrant of the first season. This was a layer of very small soil free stones (a gravel). It was cut by a shallow scoop, 757, 0.57 m long by 0.32 m wide and 0.09 m deep, on the south side of bay 785 which was filled with red ash, 756. These were sealed by a thin charcoal layer 704, 262 and thick red ash layer 680/656, 261,85. They were both thickest over bays 786 and 784, they disappeared over 785 and reappeared as thin lenses over the east edge of 743, extending to the truncated edge of the floor. In the first season this layer seemed to have occurred as a thick deposit covering the interior in both the north and east quadrant. It was labelled 16, 26, 34, and 70. 26 may represent a redeposition of the layer during one of the later reconstruction's though this is by no means certain. This layer may represent the collapse of an organic rich turf roof during a conflagration or the combustion of a layer of hay, manure and turf lying on the abandoned broch floor (see specialist analysis for further details).

This ash layer had been truncated in the centre of the broch by the construction of the revetment wall of phase 7/2 and was sealed by the walls of 7/3 and 7/4. In the centre of the broch this layer survived only underneath the kerb of the central hearth. Here, however, it was a much more mixed brown silty clay, 619 and 747. This was probably a result of disturbance during the construction of the hearth and the activity generally associated with the phase 7/3 occupation.

The only other features present in the centre of the broch which belong to this phase are pits 51, 660, 767, 774 and 783.

Feature 51, lay in the north quadrant of the first season excavation. This was a substantial feature with a lining of large packing stones, 1032. These possibly define a post pipe and indicate that this was a post hole. The fill, 52, included the distinctive green clay which derives from natural boulder clay and which was a regular occurrence in the primary phases. Layer 55 might also be part of this fill.

Pit 783 lay to the north west of the centre of the hearth. This pit was 0.80 m by 1.14 m and 0.6 m deep. It had a primary fill, 754, of green clay about 0.08 m thick, then a large rectangular slab 755 which fitted the pit with little more than centimetres to spare around most of the edge. The pit was then completely infilled with a green clay, 748, identical to that of the primary fill. Artefacts were none existent in the fill and their is no clear explanation for the purpose of this pit.

Feature 774 was unfortunately truncated by pit 584 and not fully excavated as it was only identified in the very last stages of the excavation. It was about 0.4 m across and 0.2 m deep. The fill contained several medium sized slabs but these did not stand vertically as would be expected if they were packing slabs. It was filled with a brown silty clay, 773, which contained discrete red clay lenses.

Pit 767 lay immediately to the south east of this pit under the stone kerb of the central hearth. It was 0.35 m in diameter and 0.30 m deep. It was filled with a silty clay, 749, containing lenses of ash indicating that it was closed after the fire.

Pit 660 lay on the southern edge of the central area. It was 1.5 m long, 0.9 m wide and 0.3 m deep. The fill of the pit can be divided in two. The west side of the pit was almost completely infilled with green clay with a horizontal slab at the top and a vertical block separating it from the fill to the east. On the east side this green clay acted only as a lining to the feature extending across the base and up the sides. Above the lining on the east side were three fill layers. The lowest, 662, was a thin brownish yellow lens of clay, above this 659, a brown clay lens with abundant gravel and charcoal and finally 658 a thicker layer of dark brown clay with charcoal. Both the east and west side of the pit were then sealed by an olive brown clay, 657.

Another two features, 782 and 781, were found inside the broch but these are probably natural fissures in the limestone and the few finds in the fills, 780 and 779, are likely to have been introduced by worms.

7/6 This consists of eighteen contexts:-

208, 271, 272, 512, 528, 535, 788, 1001, 1002, 1003, 1004, 1005, 1006, 1009, 1012 1024, 1053, 1054

This subdivision represents the masonry of the broch wall.

The inner revetment is numbered 1006, 1012, 512 and 272/1053 clockwise from the south east. The wall was best preserved in the south east where it stood up to 0.7 m, 4 courses, high. 512 was the west side of the broch exposed in the second season. It was separated from context 272 by a gap of around 2.4 m. The later was the best preserved part of the wall in the second season and stood 0.6 m, 4 courses, high. The inner revetment had no particular distinguishing features. The stones used were medium to large slabs and blocks with small choking stones used to level up courses.

The outer wall face was numbered 1005, 1002, 208/1054 working again clockwise from south to north. The preservation of the south and south east sections of the wall face was much better than that of the west or north. This is probably due to the fact that the stones of the outer face were always placed directly onto bedrock. In the north and south this required excavating the till and old ground surface which concealed the bedrock. Consequently the wall faced was placed in a cut and partially protected. On the west side the bedrock appears at the surface and so the wall was less protected from erosion and destruction.

In the south east the outer revetment wall, 1005, stood up to 1 m high and was constructed in a very distinctive manner. The lower section consisted of large almost square blocks surrounded by several courses of small horizontal slabs. The upper courses seemed to consist mostly of large slabs and blocks. This effect was continued to the south west where the wall, 1002, was less well preserved standing only 0.6 m high.

Most of the western section of the wall had been completely removed and only 6- 7 small stones survived of the primary course. On the north west side only nine large slabs of a basal foundation survived. Further round to the north the till was quite thick and so up to three courses, 0.75 m, of wall, 208/1054, survived in a trench. This was, however, relatively badly built and showed little sign of the alternate blocks and courses of the south side. On the north side a false revetment, 271, was also visible, particularly in the east section, immediately behind the outer revetment.

The principal feature of the wall core, 788, 1004, 1009, 1024, is the absence of material at the centre of the wall. In the north west this gap was very irregular and could simply be the result of later robbing. On the better preserved south side, however, the outer wall had a very clear inner edge which though not definitely revetted would be very difficult to explain as an accidental result of robbing. In the south east section of the wall this gap became a well defined wall chamber 1.6 m wide. The revetment wall, 1001, defining the inside edge was 4 courses, 0.7 m, high that, 1003, defining the outer edge was 3 courses, 0.37 m, high. The chamber terminated just x m short of the northern edge to the first season of excavations. The southern end is, however, less clear. A single stone may be all that it is left of the west end but the inside wall, 1001, does seem to continue beyond this point and may indicate that the chamber is a gallery.

The other distinct feature of the rubble core was a radial gap running from the gap in the inner revetment north to a distinct inward kink in the outer revetment. Just before the outer face the gap is blocked by a low slab which was burnt red and which was surrounded by a discrete layer of ash/burnt soil. It is possible that this gap marks the original entrance passage and that the burning is associated with the conflagration which marks the end of the primary occupation of the interior. The gap through the inner revetment was blocked in all subsequent reconstruction's. To the east of the gap is the one area of the wall core, 528, where there was no sign of an intra mural gallery. A solid layer of stone extended from the inner to the outer revetment defined on the west side by a line of upright stones.

7/7 This consists of six contexts:-

13, 17, 22, 23, 24, 1014

These are the layers infilling the wall chamber.

The most significant feature was a layer of slabs, 1014, which paved the northern half of the chamber. These were irregular and uneven and did not cover the floor. Overlying this was a layer of yellow silt, 17. The south side of the chamber was filled with three layers. The lowest was a layer of ash, 24, similar to that found in the interior of the broch. The layer over this (not described in the records) was a thin lens, 23, thickest against the south chamber wall. This was sealed by a yellow silt, 22, possibly equivalent to 17. The final layer, 13, associated with the use of the chamber was described as a yellow brown silty loam with some medium to large stones.

Block 11

This block isolates the features outside the broch wall in the south quad. This was divided into five blocks by R McCullogh but as there is practically no recorded stratification in this area it seems a bit excessive. Most of the features were heavily truncated. It is possible to make the case for three principal subdivisions.

11/1 This consists of seven contexts:-

44, 57, 73, 155, 156, 1039, 1051

This comprises a building defined by wall 1039 enclosing soil layer 57.

Layers 155, 156 and 73 are spatially discrete soil layers which have no stratigraphic location or interpretation and are placed here because it is the final phase.

The structures and all the other deposits in this area were overlain by rubble layer 44/1051.

11/2 This consists of four contexts:-

153, 154, 1036, 1045

This appears to be a reconstruction of the platform built around the broch. 1036 and 1045 represent the stone revetment and 154 and 153 the soil and rubble backing.

11/3 This consists of seven contexts:-

1037, 1038, 1046, 1077, 1078, 1079, 1080

1046 was a revetment wall against the broch outer face.

1038, 1078, 1080 and 1079 are stone lines which have been truncated by later activity. 1038 is the best preserved example and may surround a hearth, 1037.

These structures appear to back onto rubble layer, 1077.

Block 12

This block isolates the features outside the broch wall in the east quadrant. The area was split into five groups by R McCullogh but many contexts were not attributed to a block and some were attributed to two blocks. I propose to break the contexts down into four divisions. These essentially follow the original analysis of the stratigraphy but certain problems are noted.

12/1 This consists of 13 contexts:-

30, 36, 47, 62, 65, 66, 157, 1022, 1023, 1025, 1030, 1034, 1044

The bulk of these contexts were part of a structure which extended into the southern baulk and possibly joined up with the final structure in the south quadrant.

It includes:-

Wall 1022 which backs into rubble layer 65. This wall enclosed the northern half of the structure. the southern edge was defined by the external edge of rubble layer 1030 which appeared to be internal cobbling which lay around the edge of the house floor.

Drain 1023, cut 66 and fill 36.

Floor, paving 1025, soil layer 47.

In a similar stratigraphic position, in the south east corner, was hearth 1044, and associated ash layers 62 and 157.

Covering most of the area was rubble layer 30.

12/2 This consists of 4 contexts:-

48, 49, 50, 58

These represent a stabilisation horizon after the abandonment and decay of the underlying structures. The most widespread layer was soil 50 (which enclosed charcoal lens 58). This covered the eastern part of the area. To the west was rubble layer 49 and under the structure in 12/1 was soil layer 48.

12/3 This consists of 12 contexts:-

43, 59, 61, 64, 151, 1029, 1031, 1035, 1041, 1042, 1043, 1075

There were a disorganised mass of stone lines and occupation layers.

The main features were three stone lines, 1029, 1031, 1041, extending roughly east to west and one, 1035, extending north to south. The line that looks most like a wall 1031 but this is straight and does not seem to enclose anything.

Some indication of sequence is suggested by soil layer 61. This goes under 1041, but over 1029. It might suggest that 1041 was part of the later (12/1) structure thus making 61 equivalent to 50 (also 43 same as 49 ?).

In the centre of the area, adjacent to 1041, was a paved area, 1043, defined by a kerb, 1042, which was infilled with 64 and 59 and this is presumably a hearth.

The structure on the north side defined by stone line 1029 has hearth 1075 at its centre.

12/4 This consists of 16 contexts:-

60, 63, 75, 76, 77, 78, 79, 141, 143, 150, 1021, 1033, 1071, 1072, 1073, 1074

The main feature of this group was a wall 1033 enclosing an area on the north edge of the trench which contains hearths 1071 (fill 77) and 1072 (fill 141) and possibly floor level 76.

The floor level overlay another layer 150 in the north east corner and is overlain by 75 which appears to be an abandonment layer.

Wall 1033 was thought to be earlier than a revetment wall, 1021 (backed onto rubble layer 63), which follows the broch wall at this point, but it is possible, on examination of the plans, that 1021 was rebuilt to fit better with 1033.

In the southern part of the trench there was a damaged wall, 1074, which enclosed layers 60, 78 and 79.

In the south east corner of this area were deposits 1075 and 143, but these could belong to a later phase.

Block 8

This block isolates all the contexts that precede the construction of the broch.

There are three subdivisions.

8/1 This consists of one context:-


This is the Bronze Age cremation.

A shallow pit/scoop was dug through the natural subsoil onto the bedrock. Into this was placed a complete jar, undecorated but with three pairs of repair holes, filled almost to the top with cremated bone. A thin flat slab was then placed over the mouth of this pot to seal the contents.

This cremation was exposed when cleaning the east section of area 5 which was cut by the access road. The stratigraphy over the burial had been disturbed by the cutting of the access road and so it can not be established for certain that the burial underlay the broch wall which would have covered this area. It is, however, unlikely that the broch wall would have been removed to allow this burial to be placed in position.

8/2 This consists of 21 contexts:-

18, 25, 32, 33, 37, 40, 53, 54, 56, 88, 89, 99, 140, 142, 223, 230, 254, 270, 572, 582, 632

These are various manifestations of the undisturbed natural scattered about the hilltop.

The natural subsoil consisted of two distinct layers which overlay the pitted fractured limestone bedrock, 223. The lowest layer, 56, 89, 254, 572, was a very dark greyish or olive brown silty sand with almost no rock inclusions. The upper layer, 37, 88, 230, 270, 582, was a yellowish brown silty clay with numerous inclusions including a small number of medium sized rounded boulders.

These two layer are interpreted as a lower layer of decayed limestone and an overlying layer of glacial till.

8/3 This consists of 13 contexts:-

31, 38, 39, 41, 74, 86, 148, 267, 1019, 1020, 1026, 1027, 1028

These are the linear features and buried soil layers under the broch wall.

The old ground surface, 31, 32, 33, 86, 267, was dark yellowish brown silty clay up to 0.06 m thick at most. Large areas of this was exposed under the broch wall and on the south side in the area excavated in the first season several distinctive features were noted. These were at least four straight linear gullies, 1027, 1026, 1020 and 1028, running slightly to the east of a north to south line. The gullies were filled with vertical packing stones. It is not clear what there function might have been.

Area 2

This was an L-shaped area of surviving stratigraphy lying between house 1 and 2 and the access road. The area was examined in both seasons. In the first season the south and west sections were cleaned to vertical and drawn. In the second season the area was deturfed and the upper layers removed until it was clear that all the graves belonging to the phase 4 cemetery had been removed. Adjacent to the south and west sections there was up to 2m of stratigraphy but this dropped steeply to the north east where the phase 5 activity had truncated the stratigraphy and cut into the underlying glacial till. Over a metre of stratigraphy which includes significant structural features remains for future excavation.

The 48 contexts in this area belong to block 10 and this block consists of five subdivisions. These blocks are stratigraphically related with 10/1 overlying 10/2 and 10/2 overlying 10/3 etc. Contexts 300-320 were excavated in the second season, contexts 100-126 were excavated in the first season.

Block 10

10/1 This consists of ten contexts:-

100, 101, 111, 300, 301, 302, 303, 312, 313, 314

These contexts represent the topsoil and contexts associated with the phase 5 crofting settlement.

The topmost layer was the turf (300/100) and topsoil (301). To the north east this overlay the remains of the south west corner of the phase 5 croft. These remains comprised a wall (303) which was in a trench cut into the edge of the prehistoric mound and the underlying natural clay till. The inside of the house was covered by a layer of rubble (302) which is presumed to represent the collapsed house walls. Neither was excavated. On top of the mound to the south of the house was a layer of sea shells (111/312) and rubble (101/313/314) which appear to be the remnants of middens which originally surrounded the croft.

10/2 This consists of 17 contexts:-

102, 108, 116, 117, 122, 304, 305, 306, 307, 308, 309, 315, 316, 317, 318, 319, 320

This block consists of the graves and skeletons belonging to the Medieval cemetery of phase 4.

Seven skeletons and one pile of disarticulated bones were recognised (305, 308, 309, 117/317, 116, 102, 315). These were all aligned roughly east-west with their heads to the east. Three out of four skeletons excavated in the second season had a slight NNE-SSW orientation. Most of the graves excavated in the first season had been partly disturbed by the recent developments and were excavated in difficult circumstances. Consequently their precise orientation and any absence of bones is not likely to be a direct reflection of the original burial ritual.

Burial 309 was discovered immediately below the topsoil. A shallow grave (307) was thought to exist and the fill (308) was sampled. This skeleton was badly disturbed by truncation during the phase 5 crofting activity. The lower legs had been removed and the skull was very badly crushed. There was, however, no indication that the burial was anything but a standard extended inhumation buried soon after death with a NNE-SSW orientation.

Burial 305 was placed in a clearly defined grave (304) which had a large number of flat angular stones which may have been deliberately placed to cover the corpse. This was a simple extended inhumation on a NNE-SSW axis. The bones were well preserved and all apparently present.

Burial 117/319 was partly excavated in the first season and then finished off in the second season. It was in a clearly defined grave (320) with a relatively stone free fill (318). The bones were well preserved and the skeleton apparently complete. It is, however, possible some bones may have been lost between the two seasons due to erosion of the exposed section. The burial was aligned east-west and around the middle of the left radius and ulna was an iron bracelet. Unlike the other skeletons which had their hands by their sides this individual had 'her' hands crossed over 'her' stomach.

Burial 315 was placed in an ill-defined grave (316) with a soil fill (317) which has been sampled. The burial had been truncated below the knees by the phase 5 croft but the remaining skeleton was well preserved and undisturbed. Consequently the absence of a skull appears to be a product of ancient disturbance or as a result of pre burial decapitation. The skeleton was otherwise a straightforward extended inhumation with a NNE-SSW alignment.

The only other human remains in the area excavated in the second season consisted of a cluster of disarticulated human remains including a skull which lay in a small pile. This may represent the reburial of remains accidentally excavated when a new grave was being excavated.

All that was left of burial 122 were the legs below the knees. The rest of the body had been removed by the excavation of the access road.

Burial 116 was again badly damaged by the recent construction work. All bones below the lumbar vertebra were in situ as were both hands (only one on photo) and the left arm below the elbow. The grave was very difficult to identify and no context numbers for this and the fill were allocated.

Burial 102 was again badly disturbed. The remains consisted of the skull, upper part of the spine and rib cage and the left scapula and arm including the hand. Again no grave could be identified and no context numbers were allocated.

Burial 108 was initially recorded as coming from below 112 (block 10/3) and as can be seen in section xx it is lower than most of the other skeletons. However, the most likely interpretation is that it was in a deep grave which was not recognised due to the disturbed nature of the section. The burial was of a juvenile aligned east west with the head to the west. It is not clear from the paper records whether this was a complete or disturbed burial.

10/3 This consists of three contexts:-

112, 310, 311

These are the soil layers which were cut into by the graves of the overlying cemetery.

112 was identified as a thick layer of mid-brown silty clay with a relatively small quantity of small to medium sized stones. It would appear to reach a maximum thickness of one metre at the north west corner of the area where it lay directly over natural. 311 and 312 are the same layer excavated in the second season. The difference between the two is due to the increased frequency of stones at the top of this layer in the south west corner of the area.

This layer represents a considerable accumulation of soil before and possibly during the use of the cemetery. Much of the soil may be derived from decay and worm sorting of the remains of earlier occupation but it seems difficult to account for the considerable thickness that is present on the site solely by this criterion.

Overlying 112/311 in the south west corner of the site was a rubble layer 310. In the second season this was very clearly cut by the grave that held skeleton 305. In the first season, however, this layer was assumed to be a continuation of layer 101 which overlay the graves excavated along the south section. It also appeared to overlie burial 122 which lay within a metre of 305. There are three possible interpretations of this a. 101 was cut by all the graves and this was simply not visible due to the very disturbed nature of the stratigraphy, b. 305 is a later grave and that 101 was dumped during the use of the cemetery, c. that the subsidence of 310 into the grave 305 give the false impression that it had been cut through. I would favour either interpretation 1 and 3.

10/4 This consists of 15 contexts:-

103,105, 107, 109, 110, 113, 114, 118, 119, 120, 121, 123, 124, 125, 126

All of these contexts were only observed during the cleaning of the sections in the first season. They consist of a series of walls and occupation layers indicating considerable activity presumed to be contemporary with the LIA occupation in area 1/5.

The stratigraphy suggests a chronological span of activity. The principal feature in the south facing section was a wall, 107. This lay on two distinct orange red clay layers (109, 114) which presumably represent peat ash from hearths. The wall stood 3 courses, 0.6m high and was about 0.8 to 1m thick. This appears to be part of a structure with the interior on the south side. Two possible orange clay - ash layers were observed, 113 and 103, separated by a thick layer of silt, 105. The lowest orange clay layer (113) may be contemporary with 114 which precedes the wall but the stratigraphy was not clear. All three layers were separated from the wall by a layer of rubble and gravel, 119, and above this a dark brown silt layer 120, which makes it difficult to closely associate with the construction or use of the wall. The section to the west of the wall was not taken down as far as that to the east so it is not known what was contemporary with the wall on this side. The lowest layer observed was a layer of loose voided rubble 118 which was level with the top of the wall and possibly a result of its destruction.

In the west facing section the stratigraphy was much simpler. The most significant feature was another wall, 126, 4 courses and about 0.5m high. This was at an angle to the section and it is not clear how wide it was. Like 107 this wall overlies a patch of orange clay, 125. To the south was a soil accumulation with ash lenses, charcoal fragments and numerous large boulders. On top of this was a an orange clay - ash layer, 124, and it sealed a similar deposit, 123. There is no clear link between the stratigraphy in this section and that in the south section. It is possible that all these layers are equivalent to those described in block 10/5.

10/5 This consist of two contexts:-

104, 106, 115

These are the lowest levels examined in the south facing section.

At the base of the section is a layer of large flat slabs which formed a barrier below which excavation could not take place. This is clearly the paving of a significant structure. Above these were two thick accumulations of soil and stones, 104 and 115, both of which were heavily flecked with charcoal. 104 has a much higher quantity of large rubble and it was also distinguished by the presence of red ash lenses and large quantities of well preserved animal bone. The boundary between these two layers was almost vertical indicating that there was a deliberate cut probably with 104 the fill and 115 the truncated earlier deposit. This cut might also have truncated 109 and 114 making these two layers earlier than the fill 104.

Area 3/4/6

This area lay to the south of the main trench area 1/5. It was explored by five separate trenches of various shapes and sizes (fig 1). In the first season two small soil pits were excavated and, after cleaning and after cleaning the section of a service trench (fig 28), a 1 m square was excavated to explore a stone cist. In the second season a machine trench c 20 m long and 1.8 m wide was excavated down the west side of this area and two small trenches were hand excavated. Both these trenches were originally 2 m square and were sampled in considerable detail to examine on the formation of the middens present in this area. Removal of the upper midden in the southern trench revealed a wall and it was decided to extend the trench to examine the structure and to abandon detailed sampling of the midden. The trench ended up 4 m by 2 m with a 1.5 m extension to the west (fig 27) and incorporated the trench excavated to examine the cist in the first season.

In total approximately 51 sq. m was explored. This gives a fairly clear indication of the nature of the deposits on the hillslope to the south of the broch (but does not give an indication of the situation to the south east). A full interpretation of the partially excavated structure is, however, more difficult.

In total 31 contexts were identified in this area. These have been divided into three stratigraphically related blocks which spread across four of the trenches. No stratigraphy or archaeologically significant deposits were recovered in the machine trench.

9/1 This consists of four contexts:-

127, 400, 401, 420

These are the topsoil and turf contexts from the different trenches.

127 comes from the two sample pits of the first season. 420 came from the southern trench. In the northern trench the turf and topsoil were recorded as two layers; 400 and 401 respectively.

9/2 This consists of 12 contexts:-

128, 129, 144, 145, 146, 147, 402, 403, 421, 423, 426, 1076

This represents midden and structural features immediately underlying the topsoil.

The upper midden levels were most fully exposed in the northern trench in the second season. In this trench two distinct layers were identified. 402 was up to 0.40 m thick in the north west corner but it tailed away to the south and east and in the south west corner, where it was overlain by 0.25 m of 403, it was only 0.10 m thick. The significance of this split is, however, open to question as both layers dark brown silts with large quantities of small laminar stones. 403 had slightly more charcoal and 402 slightly more bone and pottery.

Two layers equivalent to these were identified in the first seasons sample trenches, 128 and 129. The former lay directly under the topsoil and was much stonier than 129.

In the southern trench removal of the topsoil revealed a similar dark brown stony silt, 421. Under this a line of large horizontal stones and a low bank of soil were exposed cutting across the trench from north to south. The bank of soil was approximately 1.00 m wide and seemed to overlie the large stones. It might therefore be the very badly eroded remains of a turf wall with a stone foundation. There is a slight curve on the wall which suggests it enclosed the area to the east where the cist was located. The southern part of the wall overlay a dark brown soil layer, 423, which thickened to the south. It was visible for up to 3.5 m in the section of the adjacent drain and was up to 0.40 m thick.

The first seasons excavations indicated that the cist was sealed by a layer 144/146 which appears to be equivalent to 421 and cut 147, which may be equivalent to 423. It would seem likely therefore that the cist, 1076, was contemporary with the wall. The cist was truncated to the east by the sewage trench. It consisted of one slab of the north and one slab of the south side and all of the west side. The west side was 0.7 m long and consisted of one large vertical stone and two smaller uprights. Several smaller slabs made up the paved floor of the cist. The floor of the cist was covered by a rubble layer 0.15 m thick. There were no significant finds in the cist and there is no positive evidence to choose between a burial or domestic function.

9/3 This consists of 13 contexts:-

130, 131, 132, 404, 405, 406, 422, 424, 425, 427, 430, 433, 434

This represents the lower midden layers.

In all four trenches with in situ archaeological deposits the lower levels were defined by a very distinctive red brown silt. This was labelled 404 in the northern trench of the second season and 130 and 131 in the sample squares of the first season. In the southern trench of the second season a number of very subtle changes in colour were used to distinguish a variety of different layers 422, 424, 425, 427 and 430. None of these layers were clearly defined and there seem only to indicate the variability of the midden layer. It is tempting to link this ash with the destruction of the broch. Perhaps it derives from clearing out the ash in the interior prior to the secondary occupation.

Underlying this layer were a number of thin soil layers, 405 and 406 in the northern trench, 433, 434 lay in the southern trench and may infill a large but shallow hollow. Material was very rare in these layers but it was only the presence of occasional scraps of bone and shell that stopped them from being identified as natural subsoil. They represent minimal human activity on the hillslope prior to the dumping of the red midden.